Encroachment and expansion of woody species in
the savanna areas of Masutlhe and Lekung in the
North West Province: A case study
By:
T.K.J. Sebitloane
Student number: 18029981
Previous qualification: B.Sc Hons (Biology)
Thesis submitted in partial fulfillment of the
requirements for the degree Magister Scientiae in
the Faculty of Agriculture, Science and
Technology at the Mafikeng Campus of the
North-West University
Supervisor:
Prof P.W. Malan
i
DECLARATION
I, Tshegofatso Keritebetse Joyce Sebitloane (18029981), hereby declare that the dissertation titled: The encroachment and expansion of woody species in savanna areas of Masutlhe
and Lekung in the North West Province: A case study is my own work and that it has not
previously been submitted for a degree qualification to another University.
Signature: ... Date: 2017-04-12 Tshegofatso K.J. Sebitloane
This thesis has been submitted with my approval as a university supervisor and I certify that the requirements for the applicable M.Sc. degree rules and regulations have been fulfilled.
Signed ……….. Prof. P.W. Malan (Supervisor)
Date: 2017-04-12
Signed ……….. Prof. C. Munyati (Co-Supervisor)
ii
DEDICATION
I dedicate this research to my Lord Jesus Christ, who protected, guided and strengthened me through this research and who made it possible for me to reach this level of my academia. May only His name be praised and glorified at all times.
iii
ACKNOWLEDGEMENTS
I owe lot of credit to my supervisor, Prof. P.W. Malan, for guiding, supporting and exposing me to the world of research.
I would like to thank my co-supervisor, Prof. C. Munyati, for guidance, support and introducing me to Remote Sensing.
I would also like to acknowledge the NRF (National Research Foundation) for funding this research.
I would like to acknowledge the North West University (Mafikeng Campus) for funding part of this research.
To my colleagues, Mr. Ndou and Sammy (Department of Geography and Environmental Sciences, Mafikeng Campus), for your assistance on GIS and remote sensing.
To Professor T.A. Kabanda (Department of Geography and Environmental Sciences, Mafikeng Campus), for your assistance on climatic information and formulation of graphs.
Dr. T.D. Kawadza for his assistance with the language editing of the manuscript.
A great appreciation to my mentors Harmony and Precious Monageng.
iv
LIST OF FIGURES AND TABLES
FIGURES:
CHAPTER 3: Page
Figure 3.1: Orientation of the North West Province (Department of Agriculture, Conservation,
Environment and Tourism, 2002) 25
Figure 3.2: Geographical location of the study site 27
Figure 3.3: Average minimum temperature (°C) for the years 1990-2014 29
Figure 3.4: Maximum temperature for Mafikeng (1990-2014) 29
Figure 3.5: North West Province mean annual rainfall (Department of Agriculture,
Conservation, Environment and Tourism, 2002) 31
Figure 3.6: Average rainfall of the North West Province (1990-2014) (South African Weather
Service, Station) 32
Figure 3.7: North West Province Vegetation Types (Department of Agriculture, Conservation,
Environment and Tourism, 2002) 33
Figure 3.8: Biomes in the North West Province (Department of Agriculture, Conservation,
Environment and Tourism, 2002) 33
Figure 3.9: Soil Degradation extent per magisterial district (Department of Agriculture,
Conservation, Environment and Tourism, 2002) 36
Figure 3.10: North West Province Geology (Department of Agriculture, Conservation,
Environment and Tourism, 2002) 37
Figure 3.11: A clear representation of Bush encroachment and disturbance of human presence.
Dense bushes visible in the background 37
Figure 3.12: A road and footpaths are clear evidence of land 38
Figure 3.13: Animal grazing in the area 38
Figure 3.14: Dense bushes are evidence of bush encroachment 39
Figure 3.15: Dense stands of Vachellia tortilis 40
Figure 3.16: Hard soil surface prevents grass growth. 41
CHAPTER 4:
Figure 4.1: Procedure for determining quadrant size for a height class, e.g. 1 m tall plants
(Coetzee and Gertenbach, 1977) 44
v
Figure 5.1: Woody species densities in the Benchmark 1 52
Figure 5.2: Woody plant densities in Benchmark 1 according to height classes 53
Figure 5.3: Woody species densities in Benchmark 2 55
Figure 5.4: Woody plant densities in Benchmark 2 according to height classes 55
Figure 5.5: Woody species densities in study site 1 (Lekung village) 61
Figure 5.6: Woody plant densities in study site 1 (Lekung) according to height classes 62
Figure 5.7: Woody species densities in study site 2 (Masutlhe 1) 66
Figure 5.8: Woody plant densities of study site 2 (Masutlhe 2)
according to height classes 67
Figure 5.9: Woody species densities in study site 3 (Masutlhe 2) 69
Figure 5.10: Woody plant densities in study site 3 (Masutlhe 2) according to height classes
69
CHAPTER 6:
Figure 6.1: Classified 21 August 2004 image 79
Figure 6.2: Classified 4 September 2006 image 80
Figure 6.3: Classified 14 July 2014 image 80
Figure 6.4: Overall trend in image classification class area in hectares 81
Figure 6.5: Overall trend in image classification class area in percentages 81
TABLES:
CHAPTER 4:
Table 4.1: List of materials 42
Table 4.2: List of images used 43
Table 4.3: Indication of the extent of bush encroachment TE ha-1 (Moore and Odendaal, 1987;
National Department of Agriculture, 2002) 45
CHAPTER 6:
Table 6.1: Error matrix for the 21 August 2004 SPOT image classification 82
Table 6.2: Error matrix for the 4September 2006 83
vi
LIST OF ACRONYMS
CNES: Centre National d’Etudes Spatialles
DACET: Department of Agriculture, Conservation, Environment and Tourism GIS: Geographic Information System
GPS: Global Positioning System
IPCC: Intergovernmental Panel on Climate Change NWP: North West Province
NWSEP: North West State of the Environment Report SPOT: Systeme Pour L’Observation de la Terra SAWS: South African Weather Service
TE ha-1: Tree equivalents per hectare UTM: Universal Transverse Mercator
vii
ABSTRACT
Large areas of southern Africa are affected by woody plant encroachment. The increase in the tree-grass ratio in the savannas has been attributed to the replacement of indigenous herbivores by domestic grazing animals and the intense utilization of the natural vegetation by domestic livestock. The encroachment of woody species into savanna is a global phenomenon and often has an impact in the development of the herbaceous sward. It is especially in the communally managed, such as Masutlhe and Lekung and rural areas where woody plants often develop and expand at the expense herbaceous vegetation, mainly grasses. The encroachment of woody plant species was quantified at the selected sites and compared to nearby reference sites in Masutlhe and Lekung Villages. The prominent encroaching woody species included Vachellia
tortilis, Grewia flava and Ziziphus mucronata. All selected sites, except the benchmark sites,
had woody plant densities, exceeding 2 000 TE ha-1 that will almost totally suppress grass growth. Remote Sensing techniques were used to analyse the overall trend of vegetation in the study area. High spatial resolution digital satellite images and appropriate image processing algorithms were used to facilitate monitoring of the woody encroachment. Mean Euclidean Distance Texture analysis in 3×3 moving windows enhanced woody cover. SPOT images of 2004, 2006 and 2014 were used to monitor change detection of vegetation. Land cover maps were established, comprising three classes woody vegetation, grass and bare area. Analysis of vegetation conditions trends revealed decline in grass cover with an increase in woody vegetation, especially in the villages of the study area.
viii TABLE OF CONTENTS
CHAPTER 1: Introduction
1.1 Background 1
1.2 Problem statement 3
1.3 Aims and Objectives 5
1.3.1 Aim 5 1.3.2 Objectives 5 CHAPTER 2: Literature Review 6 2.1 Savannas in general 6 2.2 Tree-grass interactions 7 2.3 Grass quality in the savanna 9 2.4 Communal farming practices 9 2.5 Tenure (Ownership and Right of Access) 10 2.6 Causes of Bush Encroachment 14 2.7 The forces known to influence the rate and pattern of
bush encroachment 19
2.8 Extent of woody plant encroachment in the Molopo Area 19
2.9 Remote sensing 20
2.9.1 Remote sensing application 20 2.9.2 Application of remote sensing in the assessment of
bush encroachment 21
2.9.3 Remote sensing vegetation and image texture analysis 22
2.10 SPOT images 22
2.11 Remote sensing and image classification 23 2.11.1 Supervised classification 23 2.11.2 Unsupervised classification 24 CHAPTER 3: Study area and climate conditions 25 3.1 Location and description 25
3.2 Climate 27
3.2.1 Climate in the savanna 27 3.2.2 Climate of the North West Province and Mafikeng 28
3.3 Rainfall 30
ix 3.4 Vegetation 32 3.5 Vegetation 34 3.5.1 Soils 34 3.5.2 Land types 35 3.5.3 Geology 35 CHAPTER 4: Methodology 42
4.1 Materials and methods 42
4.1.1 Remote sensing data 43 4.1.2 Field data and image pre-processing 43 4.1.3 Methods to quantify woody species 44
4.2 Data analysis 45
4.2.1 Ground truthing 45
4.3 Remote sensing methods 46 4.3.1 Selection of satellite images 46
4.3.2 Image processing 46
4.3.3 Pre-processing 46
4.3.4 Geometric rectification 46
4.4 Image processing 47
CHAPTER 5: Woody plant encroachment in the study area 49 5.1 Encroachment and expansion of woody plants in Masutlhe
and Lekung 49
5.1.1 Bush encroachment in the benchmark sites of Masutlhe
and Lekung 49
5.1.2 Bush encroachment in Benchmark 1 50 5.1.3 Bush encroachment in Benchmark 2 53 5.2 Bush encroachment in Lekung and Masutlhe villages 56 5.2.1 Bush encroachment in Lekung village (study site 1) 57 5.2.2 Bush encroachment in Masutlhe 1 (study site 2) 62 5.2.3 Bush encroachment in Masutlhe 2 (study site 3) 67
5.3 Conclusion 72
CHAPTER 6: Remote sensing 78
6.1 Results and discussion 78 6.1.1 Accuracy assessment 81 6.2 Discussion and conclusion 83
x
CHAPTER 7: General discussion and conclusion 85
7.1 General Discussion 86
7.2 Conclusion 87
7.3 Recommendations 88
1 CHAPTER 1 INTRODUCTION
1.1 Background
The encroachment of woody species into grasslands (bush encroachment) is a global phenomenon (Schlesinger et al., 1990; Van Auken, 2000; Roques et al., 2001; Simonson and Johnson, 2005), having first been recorded in the 1930’s and 1970’s in the savanna area of the Northern Province and KwaZulu Natal and in the 1940’s in the arid savanna of the Kalahari. Bush encroachment is caused by the increase in cover of usually indigenous trees and shrubs, usually, in response to poor management practices (Hoffman and Ashwell, 2001).
In southern Africa, the phenomenon of increasing woody plant density is commonly referred to as ‘bush encroachment’ and it involves the invasion of grasslands and the thickening of savanna (O’Connor and Crow, 1999). The grazing capacity of large areas of the South African savanna is reported to have declined as a result of being encroached by bush, often to such an extent that many previously economically viable livestock properties are now no longer viable. Removal of some or all of the woody plants will normally results in an increase of grass production and thus in grazing capacity. However, the results of woody plant removal may differ between veld types, with the outcome determined by both negative and positive responses to tree removal (Teague and Smit, 1992). Although by definition, all savannas consist of a grass and a woody component, functionally each situation is unique. Not only are there differences in physical determinants, but the biological interactions that are based on these determinants and individual species properties are unique to each spatial and temporal situation. In addition, past management practices added to the complexity, by bringing about the different kinds and degrees of modification (Teague and Smit, 1992).
The phenomenon of encroachment of woody species into grasslands, results in the shift from open grass dominated rangelands to thickets of woody plant dominated rangelands, particularly in savanna (Joubert et al., 2008; Joubert et al., 2013). According to Wiegand
et al. (2006), encroachment by woody species into grassland-dominated areas is common
2
already been completely encroached by woody species (Wiegand et al., 2006). The process is on-going in other savanna areas affecting wildlife, and the sustainability of pastoral, subsistence and commercial livestock grazing (Archer et al., 2000). Thus, encroachment has long been of concern to land managers in grasslands and savannas, but most research focused on the effects of woody plants on grass production (Archer et al., 2001), instead of the underlying ecological mechanisms driving encroachment.
Bush encroachment is the suppression of palatable grasses and herbs by encroaching woody species, often unpalatable to domestic livestock (Ward, 2005). According to Archer et al. (2000), the reduction in carrying capacity is of great significance because savannas in southern and central Africa contain a large and rapidly growing proportion and this includes the world’s human population, including many pastoralists whose livelihood is threatened by this process. Encroachment of woody plants has been among the major threats of the livelihoods of pastoralists and their ecosystem (Gemedo et al., 2006).
Bush encroachment is a dynamic process that can be rapid and, after the species has encroached over the landscape with a high abundance, it can be impossible to manage the invasion (Rejmanek and Pitcairn, 2002; Pluess et al., 2012). Removal of woody plants will normally result in an increase in grass production and thus in grazing capacity (Teague and Smit, 1992). However, the results of woody plant removal may differ between veld types (Teague and Smit, 1992). The outcome of woody plant removal is determined by both negative and positive responses to tree removal (Teague and Smit, 1992). This is because, in the savanna vegetation, the physical determinants, biological interactions and individual species properties are unique to each situation (Teague and Smit, 1992). In addition, past management practices have added to the complexity by bringing about the different kinds and degrees of modification (Teague and Smit, 1992).
Human activities have disturbed savanna ecosystems for a long time, as savannas are a resource for food and livestock breeding (Bellefontaine et al., 2000). Particularly in recent times, man has destroyed vast tracks of natural vegetation to create more arable land, often maintained in a highly unstable condition. In most situations, the determinants of savanna systems have been modified by man, either directly or indirectly. According
3
to Teague and Smit (1992), the determinants may either be primary or secondary. Primary determinants are such as climate and soil, or secondary, such as fire and the impact of herbivores (Teague and Smit, 1992). According to Britton and Sneva (1981), examples of these determinants are exclusion of fires, replacement of most of the indigenous browsers, the restriction of movement of herbivores by erection of fences and provision of artificial watering points. Many communal areas in South Africa were subjected to ‘Betterment’, which was implemented in the 1950’s (Von Maltitz, 1998). These communal areas were divided into three main resource areas:
(a) homestead, (b) fields, and
(c) rangeland (Von Maltitz, 1998).
In rural areas, communal tenure leads to overstocking and resource over-exploitation resulting in ‘tragedy of the commons’ (Hardin, 1968). The study was conducted in communal areas of Masutlhe and Lekung in the North West Province of South Africa.
1.2 Problem statement
Most researchers are of the opinion that bush encroachment is regarded as a serious threat to livestock production in southern Africa (Archer, 1990; Hudak, 1999; Shackleton and Gambiza, 2008). The selected sites of this study have been invaded by woody species, showing clear signs of being highly degraded through exploitation. Cattle grazing are the predominant land-use in the region. Dense bush, especially Vachellia tortilis and its consequent expansion, showed clear signs of interspecies competition, resulting in the rapid spread of woody species throughout the selected sites. The woody species are of little use to cattle, which are grazers, although browsing goats and sheep will utilize twigs and shoots. Grazing capacity is believed to decrease because the pioneer grasses are less notorious and less palatable than those requiring more optimum conditions (Jacobs, 2000).
Many land-use practices such as field cultivation, garden cultivation, grazing, livestock production and tree cutting in rural areas are still driven by inappropriate policy frameworks which emphasises the urgent need for local-level institutions assisting land users in sustainable land management (Von Maltitz, 2009). According to Squires et al.
4
(1992), most rangeland development projects have failed because they focused on addressing the technological aspects, without addressing socio-economic aspects.
The encroachment and expansion of woody species into grasslands results in a decrease in grass production and thus decreases the grazing capacity of the veld. This leads to a loss of soil structure, causing surface sealing, accelerated runoff, erosion and low germination of grasses (Hoffman and Ashwell, 2001). Moleele and Perkins (1998) suggested that bush encroachment takes place as a result of the exclusive use of moisture by encroachers, high soil nutrient concentrations, low fire intensity and high soil selectivity.
Over the past few decades, increasing dominance by woody encroachers and a corresponding decline in herbaceous production has been widely reported in the savannas (Ward, 2005; Scheiter and Higgins, 2009). Increases in woody plant abundance are normally accompanied by decreases in herbaceous production and undesirable shifts in composition (Archer, 1990). In southern Africa, the shift is associated with anthropogenic activities, especially high cattle densities in communal grazing areas (Van Vegten, 1981, Skarpe 1986, Ringrose et al., 1996).
The absence of fires and browsers can also increase the level of competition between woody plants by impacting biomass accumulation, survival and growth (Goncalves and Batalha, 2011). As a management tool, fire can be used to control bush encroachment and in the arid savanna, fire has the role of maintaining trees and shrubs at an adequate height and in an acceptable state for browsing animals (Trollope, 1980). Previous research, on long-term fire effects, revealed that regular burning reduced tree size regardless of the frequency of burning, but failed to eliminate woody plants or drastically alter tree diversity at the experimental site (Furley et al., 2008). Frequency of burning is an important management strategy to consider when managing woody plants (Rutherford, 1991). Fire acts predominantly by controlling the biomass of trees within the flame zone (those trees smaller than approximately 2 m), rather than as a cause of mortality. Fire browsing together acts as a powerful restriction on recruitment of trees to the mature, grass-dominating size classes (Rutherford, 1991).
5
Many studies have been performed on the above-ground competition for light, where woody species have a clear advantage, but the most intense competition takes place below- ground, where the balance between woody and herbaceous vegetation is most likely determined (Stevens and Fox, 1991). Trees have historically been viewed as superior competitors to grasses, especially in temperate zones and are widely regarded as having an impact on herbaceous production, particularly where livestock production is a primary land use (Scholes and Archer, 1997). Although grasses are better at extracting water in the upper soil layer, trees are able to persist because they have exclusive access to water in the deeper soil layers (O’Connor et al., 2014).
1.3 Aim and Objectives
1.3.1 Aim
The aim of the study was to monitor and quantify the extent of invasive woody species and the rate of expansion in the selected study sites Masutlhe and Lekung villages.
1.3.2 Objectives
* To use satellite images to detect change of woody plant succession over time * To quantify invasive tree densities in selected sites
* To evaluate the rate of expansion of woody species in selected sites as compared to a selected reference site
6 CHAPTER 2 LITERATURE REVIEW
2.1 Savannas in general
The term “savanna” once restricted to describe central South America grasslands in Spanish is now widely accepted (Edwards, 1983; Rutherford and Westfall, 1994). The Savanna Biome of southern Africa spreads from north of 22°S latitude into northern Namibia, Botswana, Mozambique and South Africa. Savannas are one of the world’s major biomes and are the dominant vegetation of Africa (Scholes and Walker, 1993). Savannas occupy 54% of southern Africa and 60% of sub-Saharan Africa (Scholes and Walker, 1993) .The vegetation of South Africa and Swaziland constitutes the southernmost extension of the most widespread biome in Africa (Mucina and Rutherford, 2006). It represents 32.8% of South Africa (399 600 km2) and 72.2% of Swaziland (12 900 km2) (Mucina and Rutherford, 2006).
Savannas are part of a continuum that includes arid shrub lands, light wooded grasslands, deciduous woodlands and dry forests (Justice et al., 1994). The Savanna Biome occupies most of the far northern part of the Northern Cape, the western and north-eastern parts of the North-West Province, extreme western parts of the Free State Province, northern Gauteng with the more isolated occurrences in the south of this province, almost the entire Limpopo Province, north western and north-eastern Mpumalanga, most of central and eastern Swaziland, low-altitude parts of the eastern seaboard, inland of the Indian Ocean Coastal Belt in Kwazulu-Natal and the Eastern Cape Provinces and with the southernmost extension Albany Thicket of the Komga to Albany District (Mucina and Rutherford, 2006). Their importance lies in the large contribution that they make to informal and subsistence economics through the supply of grazing, firewood, timber and other resources; their contribution to the formal economy as the main location for livestock and ecotourism industries and their global impact through the emissions of trace gases from fires, soils, vegetation and animals (Justice et al., 1994).
The North West Province (NWP) consists predominantly of open savanna with grazing lands evident, especially in the more arid areas where there is inadequate water for either rain fed or irrigated cultivation of crops (Hoffman and Ashwell, 2001). Woody plants invade natural grazing land on a continuous basis and pose a real threat towards productivity (Hoffman and
7
Ashwell, 2001). This implies that woody plant encroachment can be considered as an unwanted successional process. As a result, the encroachment of woody species into semi-natural grasslands has caused much concern (Partel and Helm, 2007).
Savannas are characterized by the presence of scattered trees, mostly Vachellia erioloba (Camel thorn) and Vachellia tortilis (Umbrella thorn). The shrub layer is dominated by
Senegalia mellifera (Black thorn), Grewia flava (Velvet raisin) and Tarchonantus camphoratus (Wild camphor bush) (Smit, 1999). The herbaceous sward mainly consists of
tufted perennial grasses. The most common grass is Eragrostis lehmanniana (Lehmann’s love grass). The following grass species have a more patchy distribution: Stipagrostis
uniplumis (Silky bushman grass), Heteropogon contortus (spear grass), Aristida diffusa
(iron grass), Aristida congesta (Tassel-three-awn grass), Eragrostis obtusa (dew grass),
Eragrostis superba (Saw-toothed love grass) and Enneapogon scoparius (Bottle brush
grass) (Gibbs–Russell et al., 1991; Van Oudtshoorn, 1999).
Savannas represent water-limited ecosystems. The broad-scale distributions of the main structural savanna types in southern African are highly predictable from a knowledge of the water and nutrient availability in the environment (Carter, 1994). Bush encroachment is considered a major contributor towards the occurrence and even total absence of herbaceous plants in severe cases and this is due to the ability of some plants to survive in a water-limited environment (Smit et al., 1999). It has been widely assumed that the main competition in savannas is for water. Furthermore, rains in these areas are frequently only enough to wet the soil surface. After the dominance of grasses has been broken the water is more available to shrubs and trees than before. In such a situation, woody species with shallow lateral roots have an added advantage compared to those with root systems more restricted to deep soil layers (Scholes and Archer, 1997).
2.2 Tree-grass interactions
Higgins et al. (2000), hypothesised that grass and tree coexistence is driven by the limited opportunities for tree seedlings to escape drought and the flame zone (those trees less than approximately 2 m) (Rutherford, 1981). Bush encroachment occurs due to increased tree recruitment caused by reductions in grass cover and fire intensity (Higgins et al., 2000). The classical conceptual model on tree-grass interaction in savannas is based on rooting-depth
8
separation with respect to competition for water (Walker, 1971). This hypothesis proposes that trees have roots in both the surface and deeper soil layers, while grass roots are only in the surface layer. Van Wilgen (2009), argued that the co-existence of tree-grass interaction can traditionally be explained by either equilibrium or disequilibrium models. Equilibrium models propose that tree-grass coexistence is possible, because of separation of the root niche, with trees having sole access to water in deeper soil horizons and grasses having preferential access to, and being superior competitors for water in the surface soil horizons. However, disequilibrium models propose that there is no stable equilibrium, and that frequent disturbances through the existence of competition of either grasses or trees by periodically bringing about conditions in favour of either alternative competitors (Van Wilgen, 2009).
Stuart-Hill et al. (1987), argued that the results of the negative and positive interactions on grass production is dependent on tree density. Established trees create sub-habitats which differ from the open habitat and which extend different influences on the herbaceous layer (Kennard and Walker, 1973; Tiedemann and Klemmendson, 1973; Kellman, 1979; Grossman et al., 1980; Stuart-Hill et al., 1987; Belsky et al., 1989; Smit and Swart, 1994; Smit 2004 and Smit and Rethman, 1998; 2000).
Many savanna woody plants have extensive shallow root systems but not necessarily to the exclusion of some deeper running roots. The shallow root system enables them to make use of relatively light showers when water does not penetrate far into the soil. The most often developed lateral root system, can sometimes extend 7 to 12.5 times that of the canopy radius and it includes species such as Terminalia sericea, Burkea africana and Colophospermum
mopane (Rutherford, 1980; 1983). Savanna evergreen trees tend to have deep root systems,
at least in dry savannas (Skarpe, 1996). Schulze et al. (1998), found downward transport of water in roots (inverse hydraulic lift) with water flow into deeper soil layers. The inverse hydraulic lift serves as an important mechanism to facilitate root growth through the dry soil layers underlying the upper profile where precipitation penetrates (Schulze et al., 1998). The empirical observations of root distributions (Scholes, 1988; Scholes and Walker, 1993) show that the rooting depth separation between trees and grass is real but slight. The grass-root density exceeds tree-root density to a depth of nearly 1 m. Trees and grasses both have most
9
of their roots in the top 40 cm of the soil, which is not surprising, since in dry climates water seldom penetrates below this depth in significant quantities (Scholes and Walker, 1993).
Competition experiments for water, resources, soil and moisture have shown that mature trees are competitively superior to grasses while grasses tend to out-compete immature trees (Moore et al., 1988). Furthermore, trees have both competitive and facilitative effects on grasses in their vicinity (Stuart-Hill, 1985; Barnard, 1987; Belsky et al., 1989 and Tainton, 1999). The water-use niche of grasses, both in depth and time, is completely included by the tree niche (Knoop and Walker, 1985). The water-use efficiency of grasses is not substantially or consistently higher than that of trees; hence mature trees should always out-compete grass (Knoop and Walker, 1985). This asymmetry of competitive water use creates instability in the interactions between trees and grasses. Grazing effectively weakens the suppressive effect of the grass layer on young trees in a patch of a few hectares, leading to the conversion of an open savanna patch in a tree-dominated thicket (bush encroachment). With time, tree growth and inter-tree competition will convert the bush-encroached patch to an open savanna (Scholes and Archer, 1997).
2.3 Grass quality in the savanna
A number of factors can affect grass quality and consequently, animal production in semi-arid savannas. Water availability (Milchunas et al., 1995) grazing, fire (Trollope, 1982) and soil quality (Snyman 1998, 2002) have been identified as the major factors affecting nutritional quality of grasses in semi-arid savannas. “Moderate” leaf removal by grazing animals have positive direct effects by improving photosynthetic rates, increasing availability of nutrients, reducing water stress for un-grazed plants and increasing nitrogen concentrations in some plants (Wolfson and Tainton, 1999).
2.4 Communal farming practices
Bush encroachment affects the agricultural productivity and biodiversity of 10-20 million ha of South Africa (Ward, 2005). Communal rangelands, constituting approximately 12% of the country, include the previous homelands such as Bophuthatswana (De Bruyn and Scogings, 1998), where the study sites of this research were located. Communal rangeland areas, where agriculture is largely subsistence-based, are communally-owned and managed (De Bruyn and Scogings, 1998). These areas are degraded, sustainable and
non-10
productive. Though degraded, these communal areas support a quarter of South Africa’s human population and half of the livestock population (De Bruyn and Scogings, 1998). If all these plants were condensed into a single area, they would cover the equivalent of 1.7 million hectares which is more than the total area under commercial forestry and about the size of the Gauteng Province. Biological invasion is a major threat to biodiversity and economic livelihoods in South Africa. Invasive plants cost South Africa an estimated R 6.5 billion every year (Wilson et al., 2013). According to Wilson et al., (2013), if left unmanaged, the overall impacts on ecosystem services are likely to rise by an order of magnitude.
2.5 Tenure (Ownership and Right of Access)
In many areas of the world, one of the key causes of poor land management is uncertain tenure (Bainbridge, 2007). According to Bainbridge (2007), tenure includes land use rights, land use control and other forms of access to resources. Tenure agreements may be based on ownership, agreement and custom, lease, rent or squatting. It may be limited to only the right to browse branches from individual trees in some areas (Bainbridge, 2007).
Ranching and farming without secure tenure are common in much of the world (Bainbridge, 2007). Bainbridge (2007) concluded that a rancher, farmer or herder will not invest in careful stewardship or environmental repair without secure tenure and the belief of long-term benefits which occur from current actions. According to Hoffman and Ashwell (2001), communal land tenure, refers broadly to the system affecting the approximately 13% of land that was set aside for the homelands and self-governing territories by the colonial and apartheid government. Individuals under a communal land tenure system have few rights to own or sell land (Hoffman and Ashwell, 2001). Many communal areas have been classified as degraded on the basis of the structural differences in the vegetation when compared to commercial rangelands (De Bruyn and Scogings, 1999). Free access to communal resources is a recipe for environmental disaster as communal land tenure leads to overstocking and resource over-exploitation resulting in the “tragedy of the commons’’ (Hardin, 1968). Here ‘tragedy of the commons ‘is an economic theory of a situation within a shared-resource system where individual users acting independently according to their own self-interest
11
behave contrary to the common good of all users by depleting that resource through their collective action.
According to Hoffman and Ashwell (2001), there are two distinct forms of land tenure that operate in South Africa, namely freehold and communal land tenure. Freehold tenure (commercial land tenure) essentially provides for individual or corporate ownership of a surveyed area that may be sold (Hoffman and Ashwell, 2001). Communal land tenure refers broadly to the system affecting approximately 13% of land that was set aside for the homelands and self-governing territories by colonial and apartheid governments (Hoffman and Ashwell, 2001). Political and economic conditions, demographic patterns and land use practices differ markedly in communal and commercial areas. According to Hoffman and Ashwell (2001) under communal land tenure, individuals have few rights to own or sell land which is ultimately owned by the state. On average about twice as much land is used for settlements in communal areas than in commercial farming areas (Hoffman and Ashwell, 2001).
Communal land in South Africa is characterized by continuous grazing usually with higher stocking densities than with commercial, commodity-based ranching (Scholes, 2009). According to Vetter (2013), communal rangelands are judged to be degraded based on several indices including species composition and standing biomass that compare neighbouring communal and commercial properties. According Scogings et al. (1999), communal rangelands are concentrated in the former homeland areas, which constitute about 13% of the land surface area but are home to 25% of the human population and hold about half of all livestock. Many communal areas in South Africa were subjected to the Betterment system, which was implemented in the 1950’s. There are three categories of communal rangeland in South Africa (Scogings et al., 1999)
Designated rangeland in communal areas that were established as native reserves during or before the 1913 Land Act
Rangelands that were recently commercial (freehold) farms that were transferred as part of homeland consolidation or more recent (post-1994) land redistribution
Arable lands that are either abandoned or are still in use and become a common grazing resource after harvest, with crop residues providing grazing during dry season.
12
Scholes (2009) concluded that, in many communal lands, as compared to the commercial lands, there is a strong incentive for the individual to destock and stock animals kept as assets rather than production units. Moreover, degradation in southern Africa in communal areas can be attributed in part to the inability of land users to respond to environmental clues that warn of impending changes on the land (Beinart, 2000).
The population in communal areas comprises a significant part of the communal rangeland ecosystem, making it imperative to understand how their activities influence and effect ecosystem functioning (Oba and Kaitira, 2006). Moyo et al. (2008) observed that the grazing management strategies currently employed in communal areas are principally controlled and dictated by interactions between social, ecological and institutional factors. Thus, the poorer the resource base, the greater the dependence on rangelands and the greater the mobility that lead to increased population pressure and political changes that contribute to the breakdown of this type of pastoral production (Bainbridge, 2007).
When the overall vegetation was assessed, the rangeland was not in good condition because of the heavy bush encroachment, the communal rangeland was generally in a poor condition. Thus, communal areas have more woody species than commercial areas (Terefa et al., 2007). Communal rangelands in South Africa are generally viewed as being degraded, non-sustainable and non-productive while commercial farms are perceived to be non-degraded, sustainable and productive (the very opposite to commercial farms) (Tainton, 1999; Hoffman and Ashwell, 2001). Higgins et al. (1999) concluded that communal grazing management significantly changed the composition and structure of woody plant communities.
The savanna is used largely for livestock grazing (Bagachi and Ritchie, 2010) and prolonged overgrazing is associated with land degradation (Cheng et al., 2004). In open savanna, grass biomass always exceeds tree biomass but when heavy grazing occurs, grass biomass per unit rainfall is reduced, reducing competition with trees. According to Masike and Urich (2008), livestock in Botswana is an important economic activity, practiced in communal areas and rangelands. The uncertainty of land tenure (Hoffman and Ashwell, 2001), plays a critical role in woody plant encroachment in study sites. According to Bainbridge (2007),
13
colonialism and corporate piracy often have disrupted stable, long-term tenure relationships in drylands and rights may still be uncertain or limited decades after colonial powers leave.
Overgrazing is associated with communal grazing because there is no clear land tenure or property rights agreement that make it conducive for the farmers to invest in conservation of shared rangeland (Thomas, 2008). This then, releases water and nutrient resources for trees to germinate. Dembele et al. (2006), reported that the radial gradients of plant densities, and hence cover, are related to the regeneration gradient influenced by browsing pressure and or trampling on seedlings and saplings and its dire consequence on young and mature trees.
Exploiting natural resources without thought of the future depletes what is left for short term gain without any chance of rehabilitation (Reed, 2008). Thus, past failures of development initiatives to solve the problem of environmental degradation have been attributed to the lack of consultation and involvement of the rural populations (Tainton, 1999). The degradation of land seems to be most closely related to whether a district was managed commercially or communally (Hoffman and Ashwell, 2001). In communal areas, land tenure is a complex variable, comprising several more direct influences on degradation, such as history, demography, socioeconomic conditions and land-use factors. In South Africa, communal tenure is synonymous with high population density, poverty, poor infrastructure and a strong reliance on natural resources for survival (Hoffman and Ashwell, 2001).
Communal lands have much broader livelihood strategies, of which livestock production may only be a small part (Dikeni et al., 1996). Livestock are kept for multiple reasons including draft animal power, security, milk and meat. Communal land managers aim to maximise animal numbers per area and focus on the maintenance and survival of those animals (Dikeni et al., 1996). Management of the communal systems tends to be at low cost. Communal farmers see degradation as a long-term decline in livestock survival rates that are perceived to be independent of rainfall or drought (Dikeni et al., 1996). The economic effects of such degradation in communal areas are often very slow in expressing themselves (Dikeni et al., 1996). Communal rangeland management is a challenging process, the diversity of stakeholders and their socio-economic conditions make it difficult to apply to
14
communal rangelands (Tainton, 1999). Degradation in South African communal rangelands can be attributed in part to the inability of land users to respond decisively to environmental clues which warn of impending state changes (Vetter, 2007). As population sizes near carrying capacity, increased dependence in the population growth rates decrease with increasing population size because of the effects of competition on reproductive and mortality rates. Moreover, rangelands are judged to be degraded based on several indices (species composition and standing biomass) which compare neighbouring communal and commercial properties (Vetter, 2007).
2.6 Causes of bush encroachment
Changes in woody cover have been attributed to land-use practices. The links between causes and effects of bush encroachment are still widely debated (Archer, 2005; Britz and Ward, 2007). Homewood and Rogers (1991), argued that it is difficult to generalize about causes of change in range condition because of site specific interactions among ecological features and human use.
Most often, woody plant increases have been ascribed to poor land-use practices. The general increase in savanna trees in South Africa in more recent times has been assisted by increasing CO2 concentrations (Bond et al., 2003). The causes of bush encroachment are not
simple in that bush encroachment, can occur on both heavily grazed areas as well as in areas where grazing is infrequent and light (Bond et al., 2003). To efficiently manage on-going woody plant invasion, it is necessary to monitor the species spread regularly and there is an urgent need for new techniques enabling timely, fast and precise monitoring (Hulme et al., 2009). Early and fast detection is needed to make the management cost–effective (Vila and Ibanez, 2011). It is essential to better understand the demography of thickening species by investigating aspects of their phenology and how climate, competition, fire and browsing affect large mammalian herbivores indirectly and determine changes in the structure as well as the dynamics of vegetation communities across terrestrial ecosystems (Belsky, 1994; Jachmann and Croes, 1991; Augustine and McNaughton, 1998; Harmer, 2001).
In many areas of the world, one of the key causes of poor land management is uncertain tenure (Bainbridge, 1996). Changes to the structure of an ecosystem affect both its ecological and social values and recently, tree and shrub encroachment has presented
15
numerous challenges for land managers (Noble, 1997; Bovey, 2001). Tenure includes land-use rights, land-land-use control, and other forms of access to resources (Bruce and Fortmann, 1989). A rancher, farmer or herder will not invest in careful stewardship or environmental repair without secure tenure and the belief that long-term benefits will result from current actions. These land managers usually have large areas under their control that they fail to enforce even weak environmental protections and have little funding to support management initiatives such as fencing, well building and water projects that would better protect resources (Chambers et al., 1997).
Archer et al. (1995) and Van Auken (2009) argued that the primary causes of bush encroachment are not apparent, either globally, or in southern Africa (Hoffman et al., 1999; Ward, 2005; 2010; Buitenwerf et al., 2012). The establishment of herbaceous plants can be considered as secondary succession, which is defined as succession that occurs after the destruction of part or all of the original vegetation in a site (Gabriel and Talbot, 1984). Gabriel and Talbot (1984) defined plant succession as a progressive development, finally terminating in a climax community. Climax vegetation is thus, a final stable plant community in an ecological succession which is able to reproduce itself indirectly under existing environmental conditions (Gabriel and Talbot, 1984).
Subsistence in the rural dryland areas worldwide depends on the effective and sustainable utilization of natural resources, which are increasingly threatened by land degradation (Hoffman and Ashwell, 2001). The African continent is spatially the most impacted with more than 70% of its agricultural drylands being already deserted (Hoffman and Ashwell, 2001).
Reed (2008) postulated that Bush encroachment may be caused by the changes in land-use practices rather than climate. The most important factor influencing bush encroachment was thought to be the replacement of individual browsing animals with grazers such as cattle and sheep (Hoffman and Ashwell, 2001). Overgrazing, fire frequency, soil moisture, nutrients and global warming, have also been associated with bush encroachment (Van Auken, 2009). Climate change and change in historical atmospheric carbon dioxide concentrations, ([CO2]), fire regimes, rodent populations and livestock grazing have been registered as
16
driving forces in this shift in vegetation (Archer et al., 1995; Brown et al., 1997; Weltzin et
al., 1997).
Growing populations and diminishing resources still add to the environmental stress. Factors such overgrazing, mismanagement and overexploitation of resources and ignorance are what lead reduced biodiversity and invasion and expansion of woody species (Oba et al., 2000). A review of studies from semi-arid ecosystems (Schlesinger et al., 1990; Wilson, 1998), showed that an increase in density of woody plants beyond a critical density suppresses herbaceous plant growth (Oba et al., 2000) mainly due to severe competition for available soil water. Loss of vegetation cover and lack of tree regeneration caused by heavy browsing, grazing and wood cutting are said to initiate a sequence of processes that feedback on the local climate, causing a sustained decrease in rainfall (Schlesinger et al., 1990).
Competition is a universal characteristic of all plant (and animal) communities and has a major impact on the composition and condition of these communities. The pattern of competition may also be changed by elimination of the competition of the grass layer by overgrazing. This, however, implies that more water in both deeper and surface soil becomes available for woody growth. Thus, the destruction of grasses also reduces fire damage and should lead to a regulation of woody individuals by competition between themselves (Skarpe, 1990). Competition always occurs when the demands of two or more individuals for any growth requirements are in excess of the supplying power of the environment. An inevitable consequence of the increasing density of plants in a community is that some or all of the individuals may receive an insufficient amount of a limiting resource to fulfil their needs (Tainton, 1999). The factors for which plants compete are numerous, but those for which competition is generally most intense are light, moisture and nutrients (Tainton, 1999).
For woody plants with potentially long life-spans and low post-establishment mortality rates, seedling recruitment is probably the most critical stage in the life history (Harper, 1977). Woody plants are able to store carbohydrates from the previous season and therefore expand their leaves before or immediately after the first rains (the few deep roots may assist here) (Rutherford, 1984). This allows trees several weeks of preferential resource access before grasses are able to grow enough leaf to be competitors with trees. Most tree growth
17
and shoot development takes place in the first part of the season and sometimes again at the end of the wet season (Rutherford, 1984).
The structure of the woody component of savanna is important to animals – for example tree height which determines the available browse, dense woody entanglements forming impenetrable barriers, availability of shade and protection against predators or scavengers are all critical for the animals in the area (Bothma et al., 1994). Thus, the problem of bush encroachment is a global concern because it lowers carrying capacity and this reduces livestock production (Dean and Macdonald, 1994; Jacobs, 2000; Smit, 2004). In Africa, the main encroaching species are the thorn trees (e.g. Vachellia karroo; V. reficiens, V. tortilis,
Senegalia mellifera and Dichrostachys cinerea) (Kraaij and Ward, 2006). Furthermore,
these species also tend to have high levels of phenolic compounds (e.g. tannins) in their leaves, which reduce their digestibility to livestock and wildlife. The combination of thorniness and low digestibility of Vachellia trees reduces their accessibility and natural value to consumers (Jacobs, 2000).
The difference in species composition observed between protected (fenced-off and privately owned) and unprotected (full access and utilization of resources without proper monitoring and management) savannas may be explained mostly by a combined effect on selective logging and cutting and livestock grazing. Unprotected savannas are highly subjected to human disturbances such as overexploitation, overgrazing, mismanagement total disregard of conservation strategies (Noble and Dirzo, 1997; Sagar et al., 2003).
Woody plants are thought to have a competitive edge over herbaceous plants, due to a very often extended taproot system, implying that the savanna has an inherent tendency to become increasingly woody. The negative effect of trees on grasses may result from rainfall interception, litter accumulation, shading, root competition, or a combination of these factors (Scholes and Archer, 1997). The herbaceous component of savanna communities is normally relatively shallow rooted, implying that its growth is often dependent on moisture held within that layer. The upper layers of the soil profile are available to grasses. Here, the grass has a competitive advantage over the woody species because of its fibrous root system (Scholes and Archer, 1997). Furthermore, trees are thought to have a competitive advantage in resource-rich and heterogeneous soils and less effective in resource depleted soils. Where
18
above-ground competition depends largely on grassland management (grazing, mowing, burning), below-ground competition depends largely on soil resource quantity and distribution (Partel and Helm, 2007). Tree and grasses interact through harsh root competition, but below-ground processes have been neglected in the dynamics of semi- natural grasslands. According to Partel and Helm (2007), trees and shrubs have deep root systems and they can forage over nutrient-rich and nutrient-poor patches below the ground. Herbaceous species, in contrast, have shallow root systems that cannot reach out from poor patches (Campbell et al., 1991; Grime, 1994).
The relatively high nutrient status of soil beneath, compared to between tree canopies (Bosch and Van Wyk, 1970; Kennard and Walker 1973; Tiedemann and Klemmendson 1973; Kellman 1979; Bernhard-Reversat 1982; Belsky et al., 1989; Young, 1989; Smit and Swart, 1994), would be expected to lead to a relatively higher nutrient content of the grass growing in open savanna. Topography, soil structural properties, soil moisture and nutrients all contribute to the tree-grass dynamics within savanna systems (Britz and Ward, 2007). The specific factors involved in determining the tree-to-grass ratio and bush encroachment are not well understood (Ward, 2005; Britz and Ward, 2007). Soil texture is a crucial determinant of the tree-to-grass ratio due to its effects on plant growth, soil moisture, nutrients presence and availability (Britz and Ward, 2007).
The encroachment of woody species into temperate grasslands has also been explained by increased atmospheric nitrogen pollution (Kochy and Wilson, 2001). The larger the tree, the larger the area of resource depletion and the greater its competitive effect on its neighbours. Also, a large proportion of the roots are concentrated at shallow depths (Castellanos et al., 1991; Smit and Rethman, 1998) where they would actively compete with the shallow rooted herbaceous plants.
Evidence that bush encroachment is caused by the changes in land-use practices rather than climate exist (Hoffman and Ashwell, 2001). The most important factor influencing bush encroachment is thought to be the replacement of individual browsing animals with grazers such as cattle and sheep (Hoffman and Ashwell, 2001). Thus, growing populations and diminishing resources add to environmental stress. A review of studies from semi-arid ecosystems showed that an increase in density of woody plants beyond a critical density
19
suppresses herbaceous plant growth (Oba et al., 2000), mainly due to severe competition of available soil water. Loss of vegetation cover and lack of tree regeneration caused by heavy browsing, grazing and wood cutting are said to initiate a sequence of events that feedback on local climate, causing a sustained decrease in rainfall (Schlesinger et al., 1990).
Some species may tolerate disturbances while others may disappear (Houehanou et al., 2013). The diversity of explanations for bush abundance of trees might be due to the fact that different species having varying phenologies in varying climate, soils and thus underlying mechanisms of bush abundance and thickening may differ (Joubert et al., 2008).
2.7 The forces known to influence the rate and pattern of bush encroachment
Woody encroachment drivers are many and complex (Ward, 2005; Archer, 2010). On a global and regional scale, atmospheric warming, elevated concentration of carbon dioxide and nitrogen deposition could be possible drivers (Archer, 1995; Wigley et al., 2010). According to O’Connor et al. (2014), the suppression of fire during the early twentieth century is one of the major contributing drivers of bush encroachment. O’Connor et al., (2014), argued that severe grazing by livestock or wildlife could promote bush encroachment by reducing the fuel load or by reducing grass competition. Historical changes in grazing pressure are central to understanding bush encroachment, specifically the effect of disease pandemics and the pattern on grass competition and fire suppression and in some cases seed dispersal (O’Connor et al., 2014). O’ Connor et al. (2014), however, concluded that increased atmospheric [CO2] is the major driver of bush encroachment.
2.8 Extent of woody plant encroachment in the Molopo area
Woody plant proliferation in grasslands and savannas over the past century has been widely documented and its causes debated (Archer, 1994; Archer, 1995; Van Auken, 2000). According to Molatlhegi (2008), Mogodi (2009) and Comole (2014), the extent of woody plant encroachment in the Molopo Area in the North West Province resulted in the drastic shift from a grass dominated area to a woody dominated area, changing the ecosystem completely too woodland savanna. Furthermore, the encroachment of woody species resulted in the reduction of woody biodiversity (Comole, 2014), limiting the encroaching woody species in rangelands to dense areas of Vachellia tortilis, V. hebeclada, Senegalia
20
mellifera and Dichrostachys cinerea (Molatlhegi, 2008, Mogodi, 2009, Comole, 2014) and Prosopis velutina in the riparian areas (Comole, 2014).
2.9 Remote Sensing
2.9.1 Remote sensing application
Remote Sensing is broadly defined as the science of obtaining information about an object without direct physical contact with the object (Lillesand et al., 2004). The term Remote Sensing refers to methods and techniques that make use of electromagnetic energy, such as light, heat and radio waves as the means of detecting and measuring target characteristics (Sabins, 1987).
Remote sensing has been found to be a cost effective approach to detect changes over large areas and even geographic regions and it has been of importance in monitoring the changing patterns of vegetation (Vrieling, 2006). Mapping and monitoring vegetation species in disturbed areas requires that there be extensive coverage and that quantitative, timely, accurate and regularly collected information be gathered. All these factors have made the use of remote sensing a powerful tool (Austin et al., 2009).
According to Wessels et al., (2006), remote sensing is the advanced tool for surveying and provides the synoptic view of the area. This technique offers quick and repetitive data and is accurate and potentially inexpensive and natural resource management over large areas (Wessels et al., 2006).
However, these traditional methods of mapping and monitoring vegetation have proved not to be effective to acquire vegetation cover characteristics because they are time consuming, date lagged and often too expensive (Austin et al., 2009). In contrast, remote sensing has attracted scientific awareness ensuing in the provision of varied spatial resolution imageries that are not physically feasible and cost effective but also give appropriate and precise information (Austin et al., 2009).
More recently, there is a growing demand to use information on vegetation condition in a broader regional context and to monitor achievement, and to report on progress towards
21
regional, state and national targets of vegetation condition using remote sensing (Parkes and Lyon, 2006; Neldner, 2006).
The spatial technologies of remote sensing and GIS provide possibilities for production, storage and rapid updating, of habitat maps given the threats to stability of habitats from human and natural factors (Munyati and Ratshibvumo (2010). One of the most widely used methods in vegetation type mapping using remotely sensed images is pixel-based image classification (Munyati et al., 2011).
2.9.2 Application of remote sensing in the assessment of bush encroachment
In the past, assessment of woody vegetation density and canopy cover in the vast arid to semi-arid savanna environments has been limited to analysis of field data (Adjorololo, 2008). The acquisition of field data for relatively large areas can be impractical, considering that longer time of fieldwork is required. In this respect, remote sensing applications can provide information that is quick, timely and economical for the estimation of vegetation resources over large and complex savanna environments (Adjorololo, 2008).
Moreover, remote sensing has been utilized as a primary source of spatial data to characterize patterned woody vegetation density and canopy structure in the southern African savanna ecosystems (Hudak and Weissman, 1998; Yang and Prince, 2000; Hudak and Weissman, 2001; Wessels et al., 2006). Remote sensing has also been used to develop updated vegetation maps and establish statistics about percentage woody canopy cover in semi-arid savanna environment (Stuart et al., 2006; Wessels et al., 2006).
The utility of remote sensing in the assessment of bush encroachment has been demonstrated by a number of authors. A variety of imagery has been utilized in the process, ranging from aerial photographs (e.g. Hudak and Weissman, 1998; O’Connor and Crow 1999), satellite imagery (e.g. Hudak and Weissman, 2001) combinations of aerial photographs and satellite imagery (e.g. Hudak and Weissman, 2001; Laliberte et al., 2004) to airborne hyper spectral images (e.g. Asner and Martin, 2008). On multispectral panchromatic aerial photographs, which are particularly useful in assessments involving periods predating satellite images (e.g. Hudak and Weissman; 2001; Laliberte et al., 2004), changing in image texture as woody cover increases have been shown to be key indicators of bush encroachment (Hudak and Weissman, 1998; 2001; O’Connor and Crow, 1999; Roques et al., 2001).
22
High spatial resolution is important in detecting bush encroachment on remotely sensed images (Hudak and Weissman, 2001; Laliberte et al., 2004), primarily because the encroaching bushes have small crown diameters.
2.9.3 Remote sensing of savanna vegetation and image texture analysis
Remote sensing on tree canopy cover often follows the spectral or spatial domains which are the two main approaches applied in vegetation studies (Jupp and Walker, 1997). For remote sensing investigations, the image texture response therefore, contains important information about the spatial and structural arrangement of the remotely sensed objects (Tso and Mather, 2001).
Texture information contained in remote sensing data has been very useful for a wide range of remote sensing applications, for example, to assess global land cover, image texture analysis has played an important role in the classification of vegetation communities, which have been remotely sensed (Miranda et al., 1998; Carter and Knapp, 2001).
Texture information extracted from high-resolution multispectral images has been intensively studied and considered useful for the discrimination of different land cover classes such as water bodies, urban areas and agricultural fields (Atkinson and Curran, 1997; Chica-Olmo and Abarca-Hernandez, 2000).
2.10 SPOT images
Systeme Pour L’Observation de la Terre (SPOT) is a series of Earth observation imaging satellites designed and launched by Centre National d’Etudes Spatialles (CNES) of France, with support from Sweden and Belgium. SPOT was launched in 1986, with successors following every 3/4 years. All satellites are in sun-synchronous, near polar orbits and altitudes around 850 km above the Earth (Yichum et al., 2008).
SPOT has a number of benefits over the space borne optical sensors. Its fine spatial resolution and pointable sensors are the primary sensors for its popularity. SPOT allows applications requiring fine spatial detail (such as urban mapping) to be addressed while
23
retaining the ease and timeliness advantage of satellite data. The potential implementations of SPOT data are numerous. Applications requiring frequent monitoring are well served by SPOT sensors (Yichum et al., 2008).
2.11 Remote sensing and image classification
Image classification serves a purpose to analyse the distribution of vegetation types and determine their relationships (Ndou, 2013). Mueller-Dombois and Ellenberg (1974) defined the classification of vegetation as a fundamental tool for obtaining knowledge about the vegetation cover and its relationship with the Earth’s environment. This process converts satellite data into information based on pixel values within the image. This process has made it possible for researchers to study the Earth’s surface using satellite data (Goodchild, 1994; Gao, 2009).
According to (Campbell, 2006), image classification can be applied to various sensing applications, used for image analysis and pattern recognition. Monitoring vegetation conditions is accompanied by the classification of remote sensing images (Van Til et al., 2004). For the purpose of classification and mapping of vegetation cover and scale, remotely-sensed data are used (Perumal and Bhaskaran, 2010). Yongxue et al. (2006) defined remote sensing image classification as land-use or cover class extraction from satellite imagery. Classification smooths out significant variations and simplifies images into thematic maps of land cover. According to Gibson and Power (2000), the process of classification assigns pixels which have similar spectral characteristics which are assumed to belong to the same class are then identified and assigned a unique colour. Once an image is classified, the dataset can be interrogated and the area of different classes can be set (Gibson and Power, 2000). The unsupervised and supervised class approaches can be used. Image classification is separated into a supervised and an unsupervised classification.
2.11.1 Supervised classification
Supervised classification involves three major steps, namely (1) selection or generation of training areas (2) evaluation of training signature statistics and spectral pattern and (3) classification of images (Lillesand and Keifer, 1994; Trotter, 1998). Supervised classification requires significant interaction with the analyst, to a certain category. This method involves the need for prior knowledge of the ground cover of the study site (Hasmadi
24
et al., 2009). The user (analyst) decides on the number and types of classes to be extracted.
It requires training data to generate classes defined by where the same generated training data are used to train a classified algorithm (Kamaruzaman et al., 2009).
2.11.2 Unsupervised classification
The unsupervised classification involves the examination of unknown pixels in an image and combining them into a number of classes on the basis of natural groupings or clusters present in an image (Babykalpana and Thanushkodi, 2010). This classification method requires minimal interaction with the analyst and searches for manual groups of pixels within an image (Campbell, 2006). According to Campbell (2002), unsupervised classification has the following advantages: human error is minimized, no extensive prior knowledge of the region is required and unique classes are recognized as distinct units. The disadvantages of unsupervised classification include spectrally homogeneous classes within the data that do not necessarily correspond to the information categories that are of interest to the analyst. As a result, the analyst also has limited control over the menu of classes and their specific identities (Campbell, 2002). It is on this basis that thsunsupervised classification was preferred ahead of supervised approach in the current study.
CHAPTER 3
STUDY AREA AND CLIMATE CONDITIONS 3.1 Location and description
The study area was conducted in the semi-arid Savanna Biome in the North West Province, South Africa. The North West Province (NWP) (Figure 3.1.), of South Africa is surrounded by the provinces of Gauteng, Limpopo (formerly Northern Province), the North Cape, Free State and the Republic of Botswana (Figure 3.1.). It is the 6th largest of the 9 provinces in South Africa. The North West Province is predominately rural, with 65.1% of the population living in rural areas and 34.9% in urban areas. However, the rate of urbanization is increasing, largely due to the lack of employment in rural areas (State of the Environment Report Overview, 2002).
25
Figure 3.1: Orientation of the North West Province (Department of Agriculture,
Conservation, Environment and Tourism, 2002).
The study was conducted along the selected sites along the Mafikeng-Disaneng road, approximately 27 km from Mafikeng town (now named Mahikeng) in the Lekung and Masutlhe Villages in the North West Province, South Africa (Figure 3.2). The three selected sites were located within the coordinates S 25° 47’and E 25° 21’. In general, the North West Province, is showing signs of increased soil degradation (Figure 3.9) the most severely affected areas are those that are communally managed (North West Province State of the Environment Report, 2002). The Benchmark sites (references sites), were located within the same ecological zone as the research sites (Figure 3.2).
The study area (Lekung, Masutlhe 1 and Masutlhe 2) and the reference sites (benchmark sites), were located within the vast Savanna biome, which covers large parts of southern Africa. All the areas were located in the North West Province (Figure 3.1). Each of the sites can be described as being in a poor condition because of overgrazing, overstocking of cattle, soil erosion, wood cutting and mismanagement of resources.
